Five teams in the Rust project have adopted a formal policy governing LLM usage when contributing to the rust-lang/rust monorepo. The policy was created to address three growing problems: polished LLM-generated PRs no longer signal genuine effort or understanding, LLMs exacerbate the existing review bandwidth shortage by making code easier to produce, and mechanical copy-pasting of LLM responses wastes reviewer time. The policy's core principle is that LLMs are fine for answering questions, analyzing, refining, and reviewing, but not for creating code. LLM-generated code changes are held to a higher bar than human-authored ones, requiring mandatory tests, disclosure, and prohibiting soundness-critical changes unless the author is already a domain expert. Disclosure is required for all public LLM-generated content, harassment of LLM users is prohibited, and reviewers may close non-compliant PRs without explanation. The policy is intentionally scoped to the rust-lang/rust monorepo and is designed to be revisable as the project gathers data on LLM contributions.
